6 - 1

tealackingAI and Robotics

Nov 8, 2013 (3 years and 9 months ago)

105 views





6
-

1


November 9, 2013

Chapter 6:
Synthesis: An Architecture for Computer
-
based AIS


Building on the ideas presented in the last four chapters, the purpose of this chapter is to
present an architecture for future computer
-
based

accounting systems. In Sections 6.1 and 6.2
the arguments of the last four chapters are assessed and two fundamental propositions, called
Design Principles, are formulated as the basis for further design. Section 6.4 accepts these
Design Principles as g
iven and, using Figure 6.3 as a guide, asks how data in the corporate
database might best be structured to achieve the goals of Design Principle 1. Two independent,
but not mutually exclusive, solutions are suggested. These are outlined in the latter par
t of
Section 6.4, and developed in the next four chapters of the thesis.


6.1
Design Principle 1: Inflation/Current
-
value Accounting




It is worth trying to build a cost
-
competitive, general
-
purpose, computer
-
based
accounting system that is able to gener
ate reports in inflation
-
adjusted/current
-
value terms.



Design Principle 1, stated above, was formulated after careful consideration of the arguments
presented in Chapters 2 and 3 and additional evidence, presented below, concerning interest in
inflation

accounting in times of high inflation. Note that Design Principle 1 is a general
endorsement of inflation accounting. No decision in favour of any specific form of inflation
accounting, e.g., CPP, CCA, CoCoA, or CC/CPP accounting, is implied or intended
.


Chapter 2 provided a review of normative accounting theory. First, three Objectives of




6
-

2

Accounting were identified. Second, income measurement was singled out as the most
important task undertaken by accountants. Third, the nature of the source data n
eeded for
income measurement was considered. Fourth, because of large measurement error in systems
based on independent valuation of net assets at the start and end of a period it was suggested
that income is likely to better measured, i.e., measured wit
h less error, by matching costs
against revenues. Fifth, five major systems of accounting (for asset valuation and income
measurement) were summarized in Section 2.4. No attempt was made, however, to argue that
any one of these systems was better than an
other. Consideration of the merit of the five systems
was deferred until empirical evidence from the recent price
-
change accounting experiments
had been examined in Chapter

3.


Chapter 3 reviewed the empirical literature on inflation accounting, particula
rly capital
markets research into the impact of SFAS 33 [FASB, 1979] and SSAP 16 [ASC, 1980]. The
evidence shows that market analysts in the US and the UK were unimpressed by standard
setters' attempts to create a system of inflation accounting: in essenc
e, CCA reports added
nothing to historic
-
cost
-
based GAAP reports. For many practicing and academic accountants
this evidence was a death sentence for inflation accounting.


However, it will now be argued that the empirical evidence from the early 1980s
suffers from a
data clustering problem: it only applies to relatively low levels of inflation, say, in the range
from zero to twenty percent. At higher levels of inflation the normative arguments are
supported by the empirical evidence presented below.






6
-

3

6
.1.1

Justification of Design Principle 1


This section is devoted to tests of the hypothesis that interest in inflation accounting has tended
to wax and wane with the level of inflation a country is experiencing
1
. Four cases of high to
very high inflation

in different countries this century are examined
2
. In each case there were
calls
--

not just one or two voices but a widespread clamour and in two cases government
legislation
--

for improvements to historical
-
cost accounting. It is concluded that the p
roblem
of accounting in times of changing prices is a real one, and that the lack of interest in SFAS 33
[FASB, 1979] and SSAP 16 [ASC, 1980] is due to the relatively low levels of inflation that existed
during the test period.


Once it is accepted that
inflation accounting is essential in times of hyper
-
inflation, a more
interesting question follows: What level of general inflation that must exist before inflation
accounting reports become more useful than Historical Cost? Data are limited in this area
, but
experience in Brazil, discussed below in Case 3, offers some answers to this question.






1

Tweedie and Whitt
ington (1984) offer a similar explanation of interest in inflation accounting.
They say:



During the post
-
war period, interest in inflation accounting in both the USA and the UK proceeded
continuously but at a level which tended to vary with the immediac
y of the problem, as
indicated by the rate of inflation. Mumford (1979) provides an interesting survey of the post
-
war inflation accounting debate in the UK interpreted in this manner. [1984, p.35]


2

Apart from Israel, for which data were hard to o
btain, these are the main cases of high inflation
in countries with reasonably sophisticated accounting this century. Periods of inflation prior to this
century, e.g., during the Napoleonic wars, were ruled out because of the relative unsophistication of
accounting practices of the time.





6
-

4

Case 1:

The German Hyperinflation, 1919
-
1924.


Geldmacher [1920, p.364], translated by Clarke and Dean [1989, p.295], provides a vivid
picture of the accounting p
roblems in Germany during the early stages of the inflationary
period, in 1919:



Annual profits are growing, as evidenced by many balance sheets! Mere tricks with
figures!... Commercial profit accounting has gone wrong in this day and age...
Especially s
ince November 1919 many business people have been complaining almost
ceaselessly that the money they are getting their hands on through the sale of their
goods and products is often far less than what they have to invest, in immediately
replacing the goods

sold...[then, in the context of the 3,000 marks
reported

profits of a
piano dealer, he decries] Had not the dealer sold his ten pianos, he would not have
made
any

profit, but he would now own ten pianos of the same quality [and valued at
100,000 marks], i
nstead of only one [valued at 10,000 marks].


Conditions were much worse by 1923. Sweeney [1964, p.xxi] says



the German inflation was the most extreme in history, eventually reaching a trillion to
one by the close of 1923. People there went shopping w
ith market baskets full of paper
marks, banks set up booths on street corners so that merchants could convert their cash
receipts immediately into dollars,...


In fact, inflation was such a major problem, not just for accountants, that in desperation the
G
erman government established a new bank, called the Rentenbank, in November, 1923. With
that bank came a new currency, the Rentenmark, defined as equal to the gold value of a pre
-
war mark. Described later by Beusch [1928, p.62] as the "miracle of the ren
tenmark", the
rentenmark was an indexation technique which meant that paper money could be expressed in
terms of real purchasing power, i.e., the purchasing power of the 1914 marks. This led
accountants of the day to recommend what was called the Goldmark
bilanz technique for




6
-

5

dealing with inflation; meaningful income measurement was to be effected using figures
expressed in terms of units of similar purchasing power, in this case an historical monetary
unit.


Clarke and Dean's [1989] "Conjectures on the i
nfluence of the 1920s Betriebswirtschaftslehre
3

on Sweeney's Stabilized Accounting" provides a particularly helpful review of accounting
thought in Germany during the period 1919
-
1924. Their discussion indicates that there were
many books and articles abo
ut the problem, the most influential being those by Schmalenbach
[1921], Mahlberg [1923a and b], and Schmidt [1924]
4
. The fact that (a) the government was
forced to legislate to deal with the problem of changing purchasing power, and (b) so many
accounta
nts published books and articles on inflation accounting around that time, is consistent
with the hypothesis that the high inflation of the period stimulated widespread interest in
inflation accounting.


Case 2:

Post WW2 inflation in the US, UK and Austral
ia during the 1940s and 50s, and the
explosion of normative accounting research, 1955
-
1970
.


Most of the normative accounting research from 1955 to 1970 was concerned with asset
valuation and income determination. Literally hundreds of different authors w
ere involved. It
is difficult to argue that this was all in response to the inflation of the 1940s and 50s, but many
of the major works refer to the inflation of that period as the reason for their interest in income
measurement.





3

Business Economics theorists


4

These were all cited by Sweeney [1936].





6
-

6

In the US, the year of
highest inflation this century was 1946 when the CPI rose to 18%
5
. The
following year, on 27 May, 1947, the American Institute of Accountants (later to become the
AICPA) wrote to the Rockefeller Foundation requesting a grant to set up a Study Group on
Bus
iness Income. The Study Group commissioned five monographs including the excellent
"Income Measurement in a Dynamic Economy", by Sidney Alexander [1950] (reproduced with
revisions by Solomons in Baxter and Davidson's 2nd and 3rd editions [1962 and 1977])
and
stimulated much public discussion of the income measurement problem. On page 3 of its
Report [1952], the Study Group quickly makes clear the motivation for its efforts:



The relation of changes in price levels to the determination of income is manife
stly of
crucial importance. In Exhibit II is presented a chart, made available by courtesy of
Dun & Bradstreet, Inc., showing the changes in price levels and in purchasing power
during the century ended 1950. The changes in prices that have taken place i
n the last
twenty
-
five years have emphasized the importance of this phase of the problem which
the Group undertook to study, and the present report is, therefore, largely devoted to it.
[1952, p.3]


*** Figure 6.1 goes approximately here, see next page ***


The Study Group's Exhibit II is reproduced as Figure 6.1. Amidst considerable dissension, the
Study Group's Report [1952] recommended the use of CPP accounting to



facilitate the determination of income measured in units of approximately equal
purchasi
ng power, and to provide such information wherever it is practicable to do so
as part of the material upon which the independent accountant expresses his opinion.
[Para. 12, p.105]





5

Sharpe [1985], Table 1.1, shows the increase in the CPI in the US was 18.2% in 1946, and 9.0%
in 1947.





6
-

7


Not long after, Jones [1955, 1956] provided empirical evidence of the mea
surement "error"
associated with using historical cost reports in times of inflation. His first survey [1956] was of
nine US steel companies representing 80% of the industry, from 1941
-
47. He found that
although the firms' income calculated by convention
al means had exceeded dividends by a
substantial margin each year, after adjusting for general price
-
level changes real dividends had
not been earned since 1941. The seven
-
year difference was $US750 million. The second study
[1955], commissioned by the A
AA, was of four non
-
steel companies. Again, with figures
restated for inflation (general prices), Jones found gross income for 1940
-
51 fell from 187% to
55%, that average income tax was 67% not 57%, and that dividends exceeded real income.


In the UK, the

response to post
-
WW2 inflation has been described by Tweedie and Whittington
[1984] as follows:



In the three years immediately following the Second World War, the UK, like the US,
experienced one of the sharpest inflations within living memory (as it wa
s at that time).

This naturally led to agitation from business and from others, such as academics, for
reconsideration of the traditional historical cost methods of accounting. [1984, p.44]


Responding to calls for reforms, the ICAEW issued its
Recommenda
tions of Accounting
Principles, N12
, "Rising Price Levels in Relation to Accounts", in January 1949.
Recommendation N12 rejected any form of inflation accounting. From Tweedie and
Whittington's account [1984, pp.44
-
46] there followed in rapid succession:

(i) a Committee on
the Taxation of Trading Profits (Millard Tucker Committee, appointed 1949); (ii) reports by the
Institute of Cost and Works Accountants (1952) and the Association of Certified and Corporate
Accountants (1952)
--

both of whom proposed re
placement cost valuation systems, (iii) a




6
-

8

response to those proposals from the ICAEW
Recommendations of Accounting Principles, N15
,
"Accounting in Relation to Changes in the Purchasing Power of Money" (1952), which
reiterated its belief that historical cos
ts should continue to be used as the basis for the accounts;
and (iv) a Royal Commission on the Taxation of Profits and Income, which reported in 1955.
From 1955 onwards, inflation dropped, and with it, interest in inflation accounting. In
Tweedie and W
hittington's words:



Inflation continued at a modest rate through the remainder of the 1950s and most of
the 1960s. The rates of inflation experienced during this period (ending 1968) were at
the lower end of what has been described as "the discussion ra
nge" (Burton, 1975), i.e.,
they were sufficient to stimulate discussion of the problem by academics and
thoughtful practitioners, but not significantly high to create urgent demands for reform
from companies or users of financial statements. [1984, p.46]


In Australia, Mathews and Grant [1957] made a similar study to Jones [1955, 1956] and
concluded that firms should use replacement
-
cost pricing and current cost accounting for
income measurement. Further, they recommended that company taxes should be based

on
"current income, not accounting income". Their opening paragraph [p.1] explains the
motivation for their study:



The last twenty years have been marked by one of the greatest inflations that the world
has ever known. Although price instability is no
t a new phenomenon, the inflation that
began with the 1939
-
45 War has been unique, not only in its pervasion and severity,
but also because its persistence has led to the belief that prices will never recede to
anything like their pre
-
war level.


In the f
ifteen years that followed, accounting research was almost synonymous with price
change accounting research. Notable publications, including Edwards and Bell [1961], ARS6
(written by staff of the AICPA's Accounting Research Division) [AICPA, 1963], the Am
erican




6
-

9

Accounting Association's ASOBAT report [1966], Chambers [1966], Gynther [1966], Solomons
[1966], Ijiri [1967], Thomas [1969], and Sterling [1970], were summarized in Chapter 2. It
would be overstating the case to say that
all

these studies were mot
ivated by the inflation of the
1940s and 50s. However, there were sufficient studies citing Post
-
WW2 inflation as their
motivation
6

to support the hypothesis that the high inflation of the period stimulated
widespread interest in inflation accounting.


Ca
se 3:
Hyperinflation in South America


Inflation in South America during the 1970s is another obvious place to look for problems with
historical
-
cost accounting. Referring to a delightfully absurd interaction of historical
-
cost
thinking and foreign
-
curre
ncy translation, Wolk, Francis, and Tearney [1984] describe the
"disappearing asset problem":



For example, assume an Argentine subsidiary purchased a fixed asset in December,
1974, when the Argentina peso
-
US dollar exchange rate was $0.20. The asset cos
t
20,000,000 pesos and would be translated as $4,000,000. By September, 1982, the
exchange rate was .000040; thus, the asset would be translated at $800. [p.561]


Wolk et al. go on to explain that the FASB considered indexing the original cost of the asse
t, in
this case in pesos, for the change in the general price level in Argentina, but this solution was
rejected on the grounds that it departed from historical cost. Eventually SFAS 52 [FASB, 1981]
allowed the historical US dollar cost to be reported (in

place of the functional
-
currency cost
converted through the current exchange rate) in countries where the cumulative rate of




6

For example, Edwards

and Bell [1961, p.13], ARS6 [AICPA, 1963, p.6
-
8], and Gynther [1966,
p.8].





6
-

10

inflation in the functional currency exceeded 100% over a three year period.



Table 6.1: Percent Increase in Annual CPIs for B
razil, Year 19xx

Year

70

71

72

73

74

75

76

77

78

79

CPI

incr.

23%

20%

17%

13%

25%

29%

42%

44%

41%

77%



(Source: Seed [1981], Exhibit 2
-
1, p.18)


Inflation at rates of 100% to 1,000% p.a. in countries like Argentina and Chile, is obviously so
high
7

that, as in Germany in the 1920s, accounting systems are forced to recognize its impact.
More interesting is Brazil, which experienced "double digit" inflation (as measured by the CPI)
in 19 of the 20 years, 1958
-
1977 [Tweedie and Whittington, Table

9.4, p.235], and where
historical
-
cost accounting has also been abandoned. Chapter 10 in Seed [1981, pp.131
-
146]
examines the management accounting practices of Acos
8

Villares, S.A., Sao Paulo, Brazil,
described as the largest specialty steel producer in

Brazil, with revenues of $US100 million in
1979. As shown in Table 6.1, general inflation in Brazil in 1979, the year of the study, jumped
suddenly to 77%

per annum. It might be thought that accounting in Argentina is not relevant
to, say, the US or UK,

but this is not so. The techniques Seed reports being used by Acos Villares
for management accounting include:



standard cost accounting, direct costing, flexible budgeting, responsibility reporting,
strategic planning, return
-
on investment analysis (pa
rtially based on current costs),




7

Tweedie and Whittington [1984, p.235] report 444% inflation in Argentina in 1976, and a
change in the general price index from 100 in 1970 to 253,000 by June 1980. Chile's

index over the
same period rose from 100 to 414,000, Brazil's from 100 to 1,030.


8

"Acos" is the Portuguese word for "steel".





6
-

11

funds flow management, and the evaluation of prospective capital expenditures based
on their internal rate of return [p.131].


It is clear from Seed's chapter that Acos Villares is a sophisticated user of accounting
informa
tion, and does much of its planning in constant cruzeiros. Departmental heads use
constant cruzeiros to facilitate month
-
to
-
month and year
-
to
-
year comparisons, and prices are
based on replacement costs (within the constraints of government price controls)

with all prices
quoted as subject to adjustment based on CIP
9

index changes. This evidence from Brazil is
particularly interesting because it suggests that even for countries with "only" double
-
digit
inflation, historical
-
cost accounting creates serious
difficulties. Thus accounting practices
during the South American inflation of the 1970s are also consistent with the hypothesis that
the high inflation of the period stimulated widespread interest in inflation accounting. In
Argentina's case that inter
est was sufficient to prompt government legislation.


Case 4:

The Practitioners' Response to US and UK Inflation in the mid
-
to
-
late 1970s



The final case of inflation to be investigated is that in the US and UK in the mid
-
to
-
late 1970s
(following the oil
crisis of 1973). Reading the literature written towards the end of the 1960s,
one gets the sense, e.g., from practitioners' comments at the Symposium on Asset Valuation and
Income Determination organized by Sterling [1971], that practitioners acknowledged

the
deficiencies in historic cost accounting and were ready to accept some sort of general price
-
level adjusted or replacement
-
cost accounting system
10
. Thus, when the next period of high




9

"CIP" stands for Inter
-
ministerial Price Control, the latter index being based on a sample of 36 of
the company's pro
ducts, and is computed in accordance with government regulations.


10

They would not, however, "have a bar" of net present value accounting [Staubus, 1971] or




6
-

12

inflation came along, they seemed predisposed to act quickly.



Tab
le 6.2: Percent Increase in Annual CPIs for the US and UK

Year,
19..

71

72

73

74

75

76

77

78

79

80

US

4%

3%

6%

11%

9%

6%

7%


8%

11%

14%

UK

10%

7%

9%

16%

24%

16%

16%


8%

13%

18%



(Sources: Seed [1981], Exhibit 2
-
1, p.16;



Tweedie and Whitt
ington [1984], Table 9.1, p.218)


The annual percentage increase in the CPI for the US and the UK from 1974 through 1980 is
shown in Table 6.2. Whether because of this inflation, or as a flow
-
on from the intense
academic debate of 1955
-
1970 (which itself
was motivated initially by inflation), these
relatively low levels of general inflation were enough to trigger US standards ASR 190 [SEC,
1976] and SFAS 33 [FASB, 1979]. In the UK, inflation had risen to 16% p.a. before the
Sandilands Committee was establ
ished in Autumn, 1974. However, the pressure for action
cannot have been too strong, because it took nearly five years of "due process" from the time the
Sandilands report was published until the eventual standard, SSAP

16, was issued in 1980.
One might
hazard a guess that the critical level of inflation is around 20% p.a. in the UK, and
possibly lower in the US. But such guesses are unimportant. What is important, however, is
that when the level of inflation rose high enough, the clamour for improvemen
ts to historical
-
cost accounting were loud enough for the professional accounting bodies to act. Again, this is
consistent with the hypothesis that the high inflation of the period stimulated widespread
interest in inflation accounting.







CoCoA [Chambers, 1971].





6
-

13

Conclusions on th
e need for Inflation Accounting


The empirical evidence presented above covers most of the major experiences with inflation in
the Western world this century
11
. In
each

case, as the level of inflation rose there were
widespread calls for improvements to hi
storical
-
cost accounting
12
. The German inflationary
experience is a striking example of the failure of historical
-
cost accounting and the remedy was
simple: use of a government
-
backed general
-
purchasing
-
power index. For the same reasons,
government
-
mandat
ed general
-
price
-
level
-
index adjustments are used in Brazil and Chile
today [Tweedie and Whittington, 1984, pp.235
-
241]. The US, UK, Canada, New Zealand, and
Australia have never experienced hyper
-
inflation, but whenever inflation has exceeded, say,
10% t
o 15%, for more than a few years
13
, there have been widespread calls for improved
methods of accounting for changing prices.


On the other hand, the evidence from the empirical studies reported in Chapter

3 indicates
quite strongly that stock market analyst
s in the US and UK in the early 1980's did
not

treat CCA
income as more relevant than GAAP income. It also suggests that analysts can make reasonable
estimates of CPP and even CCA income by combining information from other sources with
GAAP income.






11

Not covered above were inflation in France in the early 1920s
, inflation in Canada, New
Zealand, and Australia in the 1970s, and inflation in Israel. Response to the French inflation was
similar to that in Germany; response to Canadian, New Zealand and Australian inflation was similar
to that in the UK; and as note
d in an earlier footnote, it is hard to obtain data on Israeli inflation
accounting.


12

More formally, the null hypothesis that there was no association between inflation and calls for
reform was rejected in each case.


13

First in the 1940s and

50s and again in the mid
-
1970s.





6
-

14

How

is one to reconcile the apparently contradictory evidence that inflation accounting is both
necessary, and unnecessary? The reconciliation offered here is simply
that inflation accounting
becomes increasingly necessary as the level of inflation rises
. I
t seems that at levels of inflation
like those experienced in the US from 1980
-
1984, e.g., around 10% per annum, there may be
some glaring cases where Historical Cost accounting produces nonsense numbers (because not
all prices move in line with the genera
l level of prices), but that these problems are still not
large enough, or general enough, to be detectable in market
-
wide studies like those reviewed in
Chapter 3. At higher levels of inflation, such as those discussed in this chapter, Historical Cost
ac
counting breaks down, and some form of inflation accounting becomes absolutely necessary.


Now, the stated objective of this thesis (page 1
-
3) is to see whether cost
-
effective, computer
-
based accounting systems can be used to generate
better

accounting inf
ormation than existing
transaction processing systems. It is clear from the discussion above that at high levels of
inflation, inflation accounting information is
better

than Historical Cost information (because
HC produces nonsense), and that at low leve
ls of inflation the information content of both
systems of accounting is more
-
or
-
less
equivalent
. Since no one knows when the next bout of
high inflation will occur, it seems wise to plan now for that eventuality, and build inflation
-
tolerance into everyd
ay accounting systems. Furthermore, it seems likely that even when
general price level changes are low, there may still be times when major assets or liabilities of
some firms experience large changes in specific prices, and that it would be desirable if
the
accounting system reported the effect of those changes
14
. This provides a further reason for
building price
-
change
-
tolerant accounting systems.





14

The arguments supporting market efficiency suggest that managers and investors would learn
about such major price changes well before they were published in accounting reports. However, this
is not an argument for
excluding the information from those reports.





6
-

15


Hence Design Principle 1: It is worth trying to build a cost
-
competitive, general
-
purpose,
computer
-
based
accounting system that is able to generate reports in inflation
-
adjusted/current
-
value terms.


The reason for recommending inflation/current
-
value accounting, is that just as ships should
be designed to cope with bad weather, so accounting systems should

be designed to cope with
times of high inflation. The challenge is to design an inflation
-
accounting system that can
compete with HC in terms of simplicity, cost, reliability, auditability, and so on. The system
does not have to be as simple as HC, beca
use it potentially offers better information than an HC
system.


6.1.2

A Possible Practical System of Inflation Accounting


Without in any way limiting the generality of Design Principle 1, or the system architecture to
be developed in what follows, a conc
rete example of the sort of adjustments considered
appropriate may be helpful. Given the costs of collection and the uncertainties and difficulties
of estimating and verifying replacement costs and/or current cash equivalents, a reasonable,
low
-
cost form
of inflation accounting might be as follows:


1.

Subject to conformity to trends in specific price index series
15
, use the lower of latest
purchase cost and market prices for inventory valuation.






15

This proviso is to prevent deliberate manipulation of inventory values by buying in small lots or
at artificially high prices.





6
-

16

2.

As suggested by Carsberg [1984a], use a small number of g
overnment
-
certified price
index series to restate fixed asset costs to rough
-
and
-
ready current
-
values (with specific
revaluations encouraged for major items).


3.

Use conventional methods of depreciation.


4.

Use financial capital maintenance for income de
termination, including recognizing gains
and losses on net monetary items.


Such a system would be more robust in the face of inflation than historic cost accounting, and
not much more expensive to run. So much that is important to estimating future cash
flows is
ignored in any accounting system that any further efforts towards "accuracy" are probably
unwarranted.



6.2

Design Principle 2: The Corporate Database



Since data must be collected for the firm's special
-
purpose feeder systems for day
-
to
-
day
c
ontrol purposes, they
should

be retained in machine
-
readable form as a corporate
database for subsequent analyses.



Design Principle 2, stated above, is a conclusion drawn after considering the implications of the
findings reported in Chapters 4 and 5 fo
r the design of future computer
-
based accounting
systems. It is far less contentious than Design Principle

1, and is therefore simpler to justify.






6
-

17

6.2.1

Justification of Design Principle 2


Chapter 4, it will be recalled, examined computer
-
based accoun
ting as it is practiced today.
Present
-
day TPAS were found to be structured more
-
or
-
less as shown in Figure 4.1, with most
data being collected in a series of special
-
purpose subsystems that have been carefully tailored
to the needs of the host organizati
on. From these systems, summary journal entries were
extracted and retrieved into the firm's general ledger. Chapter 4 concluded that developments
in computer
-
based accounting systems have been aimed at assisting management by making
day
-
to
-
day recording

and reporting easier to do. As a result, computer
-
based TPAS are often
powerful, sophisticated, comprehensive systems, but they have not attempted to improve or
change the underlying system of ledger accounts with debits and credits of constant money
amo
unts that has prevailed in accounting, conceptually unchanged, for literally hundreds of
years.


Chapter 5 examined the academic literature on what was loosely termed "computer
-
based
accounting". The two most important contributions identified were Ijiri'
s Multidimensional
Accounting [1975] and McCarthy's REA meta
-
model [1982]. Ijiri's contribution was to
recognize that the money amounts debited and credited by accountants can always be viewed
as a quantity of some resource multiplied by a money value for

each unit resource. Equally
important is Ijiri's notion of causal double entry: the value assigned to the resources obtained in
an exchange is usually determined by the value of the items given up in the exchange.
McCarthy appears to have been the first

accountant to suggest that conventional double
-
entry
accounting should be regarded as an external schema, with data derived from the firm's
underlying database of facts about resources, events, and agents. His REA model may be viewed
as a development of
both Sorter's Events Accounting [1970] and Multidimensional Accounting,




6
-

18

though McCarthy explicitly sought to reject using the notion of an account as a modelling
concept for the underlying database.


Design Principle 2 is an attempt to link these two appa
rently independent streams of thought; it
relates the computer systems described in Chapter 4 to both Sorter's Events Accounting and
McCarthy's REA model discussed in Chapter 5. A problem for all those who have considered
the merits of Events Accounting h
as always been to decide what facts should be captured and
stored in the database. However, in the 20 years since Sorter's paper, computers have grown
more and more powerful, and less and less expensive. Computer
-
based information systems
have grown with

them. It is clear from Chapter 4 that a multitude of facts is presently being
fed into computer
-
based TPAS every day. It also seems likely that the trend towards more and
more complex systems will continue. Why should it stop? If more sophisticated da
ta models
can be built and maintained at reasonable cost, management will presumably be better off. The
choices about what facts should be stored in these special
-
purpose systems are dictated by the
needs of the organizations that use them. Already, syst
ems like Weber's [1986] wholesale
distribution software (discussed in Chapter 5) are much more complex, and contain far more
detail, than was ever possible in any accounting system prior to the advent of computers.


Stepping forward to, say, the year 200
0, it seems reasonable to predict that much of the data
required for accounting systems at the general ledger level will be available from the special
-
purpose feeder systems. Of course, some data will not. For instance, estimates of replacement
costs, as
set lives, net realizable values, and differential cash flows for DCF calculations, will still
have to be specially collected when and if required. However, many standard accounting
reports will probably be generated directly from data extracted from the
feeder systems. In a
sense, time, and the needs of modern organizations for information, will have rushed past




6
-

19

Events Accounting. Accountants in individual businesses will not have to decide what
information should be recorded in each event record; it wi
ll have been decided for them by
software designers. In many cases it will be a much richer picture of the activities of the firm
than Sorter [1969] would have imagined possible.


Customer name
(A)

Order No.
(E)


Date
(E)

Code

(R)


Descripti
on

Qty.

(R)


$










Total $



Figure 6.2:
R's, E's and A's of ANY Sales Order


Now, in systems with such sophisticated models of reality it seems
most unlikely

that designers
will omit key economic variables that record the R's, E's and A's of Mc
Carthy's REA model. For
instance, consider the sales orders investigated by Weber [1986]. Assuming that sales orders
relate to the future delivery of goods to different customers, the absolutely essential
characteristics of ANY sales order are details of

the goods ordered, some identifier of the sales
order (e.g., date, sales order number), and the identity of the customer. These are shown in
Figure 6.2 with the R's, E's, and A's from McCarthy's model highlighted in bold print. Therefore,
even before re
ading Weber's [1986] paper one could safely predict that
all

order entry packages
would describe the resources ordered (McCarthy's R), details of the order event such as the
date, sales order number, purchase order number, and so on (McCarthy's E), and the

customer
(McCarthy's A). That is what order
-
entry systems do. Some systems do much more; none do
less. In other words, with the exception of McCarthy's duality relationship (discussed in




6
-

20

Chapter 5), it should ALWAYS be possible (trivially so) to derive

an REA view from an order
-
entry database. Weber's paper may be viewed as confirmation of this prediction.


It is therefore predicted that as more and more sophisticated special
-
purpose computer
information systems are built, primarily for short
-
term co
ntrol purposes, NONE will omit the
key relevant economic variables, i.e., the R's, E's, and A's of McCarthy's model
. The systems
would simply be unworkable if these key economic variables were not captured. Furthermore,
because these systems must capture

these R's, E's, and A's, their databases will provide an
excellent source of data for income measurement. Both Sorter's and McCarthy's goals for
Events Accounting will be met if accounting systems can scan through the facts stored by these
special
-
purpos
e systems, extracting and reorganizing data for summary accounting reports.


Design Principle 2 follows: Since data must be collected for the firm's special
-
purpose feeder
systems for day
-
to
-
day control purposes, they
should

be retained in machine
-
reada
ble form as a

corporate database for subsequent analyses.


For most firms, detailed data describing almost all relevant economic events are or will be
available in machine readable form as they are processed through the relevant subsystems. The
only add
itional costs associated with Design Principle 2 are for organizing and storing data
until they are discarded. As storage costs continue to fall, it seems likely that firms will find it
useful to store all transaction information for quite long times, e.g
., three to five years, possibly
more. If storage costs prove excessive, it should still be possible to retain meaningful
aggregations of more detailed data for analysis purposes.






6
-

21

6.2.2

Summary: Design Principle 2 (the corporate database)


As more and

more special
-
purpose systems are developed, primarily for day
-
to
-
day control and
planning purposes, more and more facts about the firm will be available in machine
-
readable
form at low cost. There is little point trying to predict what additional data fi
rms will store, and
different industries may have quite different needs. But their common need is for income and
wealth measures, and it may safely be predicted that almost all the transaction data (the R's, E's,
and A's) for these measures will be captur
ed by each firm's special
-
purpose subsystems. The
prime recommendation presented in this section, Section 6.2, is that the facts from all these
data
-
rich subsystems should be saved in some sort of corporate database for analysis purposes,
including the ge
neration of inflation
-
tolerant income and wealth measures. It seems reasonable
to predict that future accountants will spend less time preparing routine reports, and more time
making special
-
purpose analyses of data extracted from these corporate database
s.


6.3

Consistency of the above two Design Principles with the Objectives of Accounting
identified in Chapter 2


It would be unwise to proceed without checking to ensure that Design Principles 1 and 2 are
consistent with the three Objectives of Accountin
g identified in Chapter 2. Such checks are
rather mechanical, so they have been relegated to Appendix 6.1. Briefly, neither Design
Principle is
inconsistent

with the three Objectives. The benefits of Design Principle 1 will only
become apparent in times

of high inflation, or more generally, large price changes. During
such times, Design Principle 1 assists with all three objectives (i.e., for management
information, earning power predictions, and accountability reporting). The major potential
benefits
of Design Principle 2 are for management analyses, and as a source of data for the




6
-

22

generation of inflation
-
tolerant financial accounting reports (i.e., as support for Design
Principle 1).


6.4
An Architecture for Computer
-
based AIS


This section marks t
he turning point in this thesis. The important accounting decisions about
the goals for computer
-
based accounting systems have been made. The task now is to
transform the two general Design Principles from Section 6.1 and 6.2 into an architecture for
the

design of future computer
-
based accounting systems. From this point on, the thesis
becomes increasingly technical.


The first Principle from Section 6.1 requires that future accounting systems be inflation
-
tolerant; they should not generate misleading in
formation in times of inflation. The second
Principle requires that whenever new computer systems are being developed for day
-
to
-
day
management and control, accountants should ensure that sufficient data may be extracted from
these systems for inflation
-
t
olerant income and wealth measures to be generated. It is
considered highly likely that these special
-
purpose systems will contain, or have electronic
access to, almost all relevant facts necessary for income and wealth measurement.


The design problem no
w, portrayed in Figure 6.3(a), is to take the databases of facts captured
and recorded by the special
-
purpose subsystems and to use them to assist with the generation of
current
-
value accounting reports. How can this be done in as general a way as possibl
e?


*** Figure 6.3 goes approximately here, see next page ***






6
-

23

As an anchor point, Figure 6.3(b), based on Figure 4.1, shows the method used in computer
-
based accounting systems at present. General Ledger journal entries are either generated by, or
extrac
ted from, the various subsystems, then fed to a general ledger package. Note that the
interface programs shown in Figure 6.3(b) may be quite simple. For instance, journal entries
from Accounts Payable, Accounts Receivable, and Inventory Control in the AC
CPAC system
reviewed in Section 4.2 are generated automatically by those systems, for retrieval directly into
ACCPAC's GL system, without conversion.


There are two weaknesses with the system outlined in Figure 6.3(b). First, it may be difficult to
use t
he data in the subsystem databases effectively. Even when interface programs exist the
information they provide is likely to be very restricted. One has to rely on all the different
subsystems and interface programs generating
appropriate

journal entries
, and changes in
accounting policy, even simple changes like a revision to the chart of accounts for the general
ledger, will have implications for many different subsystems. Second, conventional general
ledger systems are decidedly intolerant of inflatio
n. The balance in a general ledger account is
constant, in money terms, until it is changed by a journal entry. It is then constant again, until
the next journal entry. However, market values are constantly changing, particularly in times
of inflation,
so accounting information may be more useful if values of assets and liabilities, and
even balances in revenue and expense accounts are continuously restated.


To overcome these two quite independent problems it is proposed that future computer
-
based accou
nting systems should be designed as shown in Figure 6.3(c). Figure 6.3(c) is the
proposed architecture for computer
-
based accounting systems, the major research
contribution of this thesis.

The novel features of Figure 6.3(c) are first, Formula Accountin
g,
and second, the Interpreter combined with the Resources and Exchange Events (REE) model.




6
-

24

The Interpreter and Formula Accounting could be implemented
quite independently

of one
another, but both have the potential to make significant improvements to the

present state of
computer
-
based accounting.


To emphasize this independence, there are two paths from the subsystem databases in Figure
6.3(c) to the Formula Accounting general ledger. In the upper path it is assumed that interface
programs similar to
those used in Figure 6.3(b) generate the FA journal entries. In the lower
path, a database schema called the REE model and a program called an Interpreter have been
interposed between the subsystem databases and the FA general ledger. Because Formula
Acc
ounting is common to both paths it will be discussed first.


6.4.1

Formula Accounting


In a Formula Accounting system, the journal entries and ledger balances
contain formulae
, not
just constants. Evaluation of a formula occurs whenever an account is used

for reporting, or is
displayed on the screen. This is exactly analogous to the way formulae in spreadsheets are
displayed as values when viewed on the screen. In Figure 6.3(c) a Formula Accounting general
ledger replaces the conventional general ledger
from Figure 6.3(b). Data flows from the
subsystems to the Formula Accounting general ledger have also been transformed from
conventional journal entries in Figure 6.3(b) into Formula Accounting journal entries (FAJEs) in
Figure 6.3(c).


Formula Account
ing (FA) is primarily intended to solve the problems of using inflation
accounting in a conventional general ledger framework, but it is also useful for HC accounting.

The examples in the next chapter, Chapter 7, illustrate both HC/GAAP and CC/CPP




6
-

25

account
ing journal entries. An implementation of Formula Accounting, which amplifies the
examples in Chapter 7, is presented in Chapter 8.


6.4.2

The Interpreter


If an organization chose to generate its Formula Accounting journal entries using the
lower

path of

the Figure 6.3(c) architecture it would need an REE data model and an Interpreter. As
envisaged in the proposed architecture, the Interpreter's job would be to make important
accounting decisions like those made by a human accountant in today's systems,
e.g., how assets
should be valued, and how the real world is to be modelled in terms of journal entries. Because
of the importance of valuation and interpretation of accounting rules in income measurement,
there may be advantages if all these accounting d
ecisions are made in a separate module, quite
distinct from any of the subsystems feeding it with information. Thus, in the lower path
proposal, the Interpreter, not the interface programs, contains the valuation and accounting
rules required to generate
the FA journal entries.


The two main reasons in favour of a separate Interpreter approach are:


(a)

Modular, loosely coupled systems, are usually more robust and error
-
free than tightly
-
coupled systems. Presumably specialist subsystems will be rewritte
n, replaced, split in
half, integrated, and added to, at different times. In addition, the accounting system itself
will be probably need to change due to changing accounting standards. In this ever
-
changing world it may be advantageous to concentrate re
sponsibility for accounting rules,
including knowledge of the general ledger chart of accounts, in one module, not to spread
that responsibility over the various subsystems.





6
-

26


(b)

Not all valuation information is available in the subsystems anyway. Data on

current
replacement costs and net realizable values (if required) has to be collected externally.
This is indicated by the "special information" feeding up into the Interpreter in Figure
6.3(c). Again, this points to the possible merit of centralizing g
eneration of Formula
Accounting journal entries.


Details of a prototype implementation of an Interpreter are given in Chapter 10.


6.4.3

The Resource and Exchange Events accounting model


Potentially the greatest difficulty with the lower path proposal in

Figure 6.3(c) is the definition
of the corporate database that provides data for the Interpreter. All that the Interpreter "sees"
when it looks for data is a database in REE format. Data of the type listed in Table 2.1 must be
collected from the various

interface programs.


Implicit in the lower path in Figure 6.3(c) are a number of new interface programs that provide
the Interpreter with less
-
processed information about economic events than the journal entries
from interface programs 1, 2, etc. in Figur
e 6.3(b). The challenge is to find a data model that is
as general as conventional double
-
entry accounting, yet with a sufficiently rich description of
the economic resources and events that have occurred for inflation
-
adjusted income and wealth
measures
to be generated automatically. Whether this corporate database is implemented as a
physical or virtual database is unimportant. Figure 6.3(c) is drawn for the case where it is a
virtual database, a view derived from the subsystem databases. However, if
the subsystem
databases were automatically purged of old transaction data, the relevant information would




6
-

27

have to be saved in a real database.


The most obvious candidate to turn to for a general data structure for the corporate database is
McCarthy's [1
982] REA model, previously discussed in Chapter 5. Could a series of REA views
derived from the subsystem databases be used as a sufficiently complete, and general, source of
data for the Interpreter? The answer is that it does seem that something
simila
r

to a Resource
-
Events
-
Agents framework would be suitable for the detailed historical record requirements of
types (i) and (ii) from Table 2.1. For instance, Ijiri's Multi
-
dimensional Accounting [1975] is
based around the assumption that details of a firm
's sales and purchases (including purchases
of services like labour, advertising, and rent), borrowings, lendings, and capital transactions,
can all be thought of as events where resources are exchanged. This can be extended to a
manufacturing firm, by vi
ewing the various inputs being exchanged for the outputs of each
production process.


However, as discussed in Chapter 5, there are some problems with using REA as the interface
model. For this reason a new data model, similar in scope to McCarthy's REA,

was developed
for use with the prototype Interpreter. That model, called the Resources and Exchange Events
model (REE) is presented in detail in Chapter 9. The REE model refines and extends REA in the
two following areas:


(a)

REE uses "accounts" as a m
odelling primitive. As discussed in Chapter

5, McCarthy was
adamant that REA
-
type models should not include "elements of double
-
entry bookkeeping
such as debits, credits and accounts" [1982, pp.559
-
560]. He therefore developed his
causal chain concept of

duality in an attempt to link events like sales and cash receipts,
and so avoid concepts like accounts receivable. However, when Ijiri's causal double
-
entry




6
-

28

approach is used as the theoretical basis for asset valuation on date of exchange, an
account
-
bas
ed approach, like his Multi
-
dimensional accounting, is actually quite a useful
way of describing relevant economic events. In addition, an account
-
based approach
makes it relatively simple to extract resource
-
event information from subsystems like
Account
s Payable and Accounts Receivable that use an explicit account
-
based framework
for recording economic events. This is an important consideration because the Accounts
Payable subsystem, in particular, is the source of many journal entries in most present
-
d
ay
accounting systems.


(b)

Resource description is essential for the Interpreter. McCarthy's REA paper is almost silent
about how resources change value over time. In his Figure 10 [p.574] he shows
Depreciation, and Advertising Service Consumption, as

two examples of economic events,
in an awkward construction he calls "macro
-
level duality". The position taken in this
thesis is that changes in value over time are
not

economic events (the only "events"
modelled in REE are "exchange events"). Instead,
all resources are assumed to change in
value over time, and resource
-
value modelling in REE enables the Interpreter to meet the
asset valuation requirements of Table

2.1. All resources in the REE model are potentially
linkable to a price index series of s
ome kind. In the case of depreciable
-
type assets, the
resource description also includes information about the life of the asset and the expected
pattern of future benefits.


6.4.4

Multiview Accounting


The final observation to be made about the lower pat
h of the architecture in Figure 6.3(c) is its
potential for generating alternative views of economic reality. If the special
-
purpose




6
-

29

subsystems continue to evolve as predicted, almost all journal entries posted to the Formula
Accounting general ledger wil
l be generated automatically from data extracted from them. It
would therefore be relatively painless to maintain a number of different general ledgers, each
using a different set of accounting rules. A system with more than one set of "books" may be
cal
led a "Multiview Accounting system" since it provides multiple accounting views of the
underlying economic reality.


If current values for assets, liabilities, revenues, and expenses, were used in the firm's normal
books, but not allowed for taxation purpo
ses, there would be a real need for two full sets of
books. Figure 6.3(c) could therefore be modified to include two separate Interpreters and two
general ledgers (or one Interpreter programmed to feed two separate general ledgers). One
ledger would be f
or historical cost tax accounting, the other for current
-
value accounting. Of
course, many organizations today say they keep two sets of books, the normal books, and the
tax books, but this is usually just a matter of duplicate records with different depr
eciation rates
in the Fixed Asset system. Unless the taxation law is changed, any move to, say, CC/CPP
accounting for normal reporting purposes would make Multiview Accounting essential.


Aside: In present
-
day computer
-
based accounting systems, like thos
e portrayed in Figure
6.3(b) (not (c)), the idea of maintaining two full sets of books would probably not be feasible. It
takes so much highly skilled management effort to ensure that one system works that proposals
for a second system would be resisted s
trongly. It is this need for constant managerial input
that makes present
-
day accounting systems so difficult to use. Yet as systems become more
automated, and more and more routine accounting is taken over by programs like Interpreters,
it becomes quite

feasible to consider multiple sets of accounting "books".






6
-

30

Multiview Accounting is discussed in more depth in Chapter

10, the chapter on the Interpreter.


6.5

Chapter Summary


This chapter is both the culmination of Chapters 2 through 5, and the foundat
ion for the
remainder of the thesis. The decisions made in this chapter determine the direction for the
remainder of the thesis. In Sections 6.1 and 6.2, after careful consideration of the implications
of Chapters 2 through 5, two Design Principles for t
he design of future computer
-
based
accounting were enunciated and justified. In Section 6.4, these Principles were represented
graphically in Figure 6.3, and the requirements for the computer
-
based accounting system of
the future began to emerge. Figure
6.3(c) shows the outcome of the design choices considered
in this chapter. It is the proposed "Architecture for Computer
-
based Accounting Systems" that
motivated the title for this thesis.


There are actually two alternative system designs within the one
architecture shown in Figure
6.3(c):


(a)

In the upper path in Figure 6.3(c), the separate feeder systems generate Formula
Accounting entries either themselves or via the interface programs. The upper path avoids
the need for an Interpreter and data trans
formation into REE format.


(b)

In the lower path in Figure 6.3(c), Formula Accounting journal entries are generated by
the combination of the REE model and the Interpreter.


Both systems satisfy the constraints imposed by the two Design Principles, i.
e., they both exploit




6
-

31

the availability of data in the various specialized feeder systems (Design Principle 2), and they
both meet the need for inflation
-
adjusted/current
-
value accounts (Design Principle 1).


The lower path appears more complex than the u
pper path because it requires definition of the
REE data model and construction of the Interpreter. However, in the upper path Interpreter
-
like functions would still have to be built into the subsystems or their interface programs.
(Otherwise they could
not generate the appropriate Formula Accounting journal entries.)
Moreover, since the Formula Accounting journal entries from either path are the same, the data
requirements of both paths must be identical. For this reason, the remainder of this thesis w
ill
be devoted to an investigation of the
lower

path in Figure 6.3(c). This has the advantage that
the data requirements, and the rules for generating FA journal entries, will be made quite
explicit.


In the remainder of the thesis the components of the a
rchitecture depicted in Figure 6.3(c) are
considered in more depth. Chapter

7 explains how a Formula Accounting system would work.

Chapter 8 reinforces these examples by describing a prototype implementation of Formula
Accounting in a programming languag
e called Prolog. Chapter 9 introduces the Resource and
Exchange Events model. This model is much more complex than the simple debits
-
and
-
credits
model used in conventional accounting systems. It has to be complex because it is attempting
to store suffic
ient information about the real world for the Interpreter to replace most of the
labour
-
intensive human judgement that goes into accounting systems today. Chapter 10
develops a set of rules to be encoded in the Interpreter. The Interpreter is to act like

an expert
system, generating Formula Accounting journal entries from data extracted from the
subsystems.






6
-

32

(8,400 words)





6
-

33

Appendix 6.1:

Reconciling the Two Design Principles from this Chapter with the Objectives of
Accounting from Chapter 2





1.

To provide management with information useful for planning and control.




2.

To provide external users with information useful for predicting the future cash flows and the
earning powe
r of a firm.



3.

To provide accountability information useful for contract monitoring.



Table A6.1
The Objectives of Accounting

(from Chapter 2)



Table A6.1 reproduces the three main Objectives of Accounting identified in Chapter

2. To
ensure that D
esign Principles 1 and 2 are consistent with these Objectives, their implications for

each Objective will now be considered, briefly, in turn.


Design Principle 1 and the Objectives of Accounting


(a)

Design Principle 1

(Inflation accounting) and
Objecti
ve 1

(To provide management with
information useful for planning and control).



One of the fundamental questions a manager must ask time and time again is "Is my firm better
off if I sell item X at price P or not?" To answer this question the manager mu
st have some way
of estimating the value to the firm of retaining X. At this micro level, deprival value seems the
most relevant of all the valuation systems reviewed in Chapter 2. As an approximation to
deprival value, it seems reasonable to request tha
t an accounting system should be able to
provide gross profit reporting based on current replacement costs (as, to some extent, reports
based on LIFO inventory valuation do).


After many such micro decisions have been made, it is also reasonable for an o
wner or manager
to ask about the overall state of affairs. How much better off is the firm as a result of this
period's activities? This question, and approach, seem to be the basis for the accounting
practice of matching expenses against revenue. It wa
s shown in the cases above that in times of
high inflation, Current Cost Income from Continuing Operations (CCIFCO) or even Constant
Purchasing Power Income from Continuing Operations (CPPIFCO),
should

be a more useful
performance indicator than historical

cost
16
. For example, it is apparent from the Acos Villares
example in Case 3, Section 6.1.1, that when inflation was high, some form of inflation
accounting became necessary for management planning and control.


Evidently, therefore, Design Principle 1
is not inconsistent with Objective 1. Its importance,




16

As with all accounting systems, CCIFCO ignores changes in probable fu
ture cash flows. In a
stable, mature corporation this may not be a problem. In a small growth company, for example
Sarich Corp., measuring CCIFCO may be like measuring the tail of the donkey and ignoring the body.

Managers would have to rely on other me
asures of success.





6
-

34

compared with the use of Historical Cost figures, will depend on the prevailing level of
inflation.


(b)

Design Principle 1

(Inflation accounting) and
Objective 2

(To provide external users with
info
rmation useful for predicting the future cash flows and the earning power of a firm).


Much of the debate in the accounting literature for the last 30 years has been over the impact of
inflation accounting on procedures for estimating future cash flows and

earning power.
Obviously, therefore, it cannot be resolved here. But in the context of very high inflation, the
issue is considerably clearer. It
is

clear that:


(i)

Irrespective of the level of inflation, in an active securities market, market efficie
ncy, even
with its limitations (Section 3.1), ensures that disclosure is more important than any single
income measure or asset valuation technique [Beaver, 1981]. (For instance, one can be
reasonably confident that even if inflation were running as high
as, say, 40% p.a., analysts
with access only to historic cost accounting reports, would be able to make a reasonable
job of estimating a firm's income in current terms.)


(ii)

As shown by the piano retailing example in the case in Section 6.1, in times of
very high
inflation, historical cost income measures are meaningless, whereas an inflation
-
adjusted
income measure, e.g., CPP or CCA income from continuing operations, is likely to be much
more useful as an indicator of future earning power.


(iii)

In time
s of very high inflation, inflation
-
adjusted asset values (CPP, CCA, or CoCoA) are
more relevant to almost any decision than "vanishing" HC asset values.


However, under relatively low inflation in the US and UK in the early 1980s, the stock markets
of t
he world behaved as if HC earnings and CCA earnings had very similar information content.

Thus Design Principle 1
is

consistent with Objective 2. Its importance will, however, depend
on the prevailing level of inflation.


(c)

Design Principle 1

(Inflatio
n accounting) and
Objective 3

(To provide accountability
information useful for contract monitoring).


Much detailed accountability information is captured and stored by feeder systems such as
accounts receivable, accounts payable, and inventory control.
Presumably such accountability
information does not need to be expressed in inflation
-
adjusted terms.


However, some accountability and contract monitoring goals need summary data, and GAAP
figures are often used in this role. The problem here is that man
y of one's normal assumptions
about the usefulness of GAAP summary figures are shattered if one imagines accounting in a
time of 40% or higher annual inflation. For instance, consider the case of debt covenants. How
are concepts like principal and intere
st defined if inflationary expectations (and interest rates)
for next year are somewhere between 20% and 70%? Will the tax laws allow firms to treat
interest payments at 50% p.a. as fully tax deductible

expenses? If the tax laws are defined relative to
an inflation
-
adjusted government bond, like
the Brazilian ORTN, then one has inflation accounting by default. Concerning executive
compensation plans, in times of high inflation it would seem rational for shareholders to prefer
inflation
-
adjusted measures

of income to historic cost income as the basis of rewarding




6
-

35

management for its performance.


Thus Design Principle 1 is not inconsistent with the accountability goal of Objective

3, and it
could positively assist with contract monitoring.



Design Prin
ciple 2 and the Objectives of Accounting


(a)

Design Principle 2

(the corporate database) and
Objective 1

(To provide management with
information useful for planning and control).


There are two reasons for believing that a corporate database would be desi
rable for
management accounting. First, the experience with market
-
based accounting research has
been that when a database of facts was collected and subjected to sophisticated analysis, some
most important facts emerged. This was certainly the case with

the CRSP
17

database at the
University of Chicago in the late 1960's, and all the studies examined in Section 3.2. Design
Principle 2 may therefore be assist with the discovery of hitherto unknown relationships to
improve management planning and control.


Second, crude techniques such as allocating overhead on single variables, e.g., direct labour,
have been strongly criticized by authors such as Johnson and Kaplan [1987] and Foster and
Horngren [1988]. One of the strongest messages to come from Foster's 1
988 seminar at the
Melbourne Graduate School of Management was that if accountants are going to meet the
challenge of new manufacturing techniques, such as Just
-
in
-
Time and Flexible Manufacturing
Systems, they need to analyze data in new ways. Management
accountants of the future using
sophisticated analytical techniques like input
-
output analysis and multiple OLS regression will
need data. It seems highly likely that the data they will need for such analyses are already being
captured and discarded by pr
oduction planning systems in many firms around the world.
Evidently, therefore, Design Principle 2 is not inconsistent with Objective 1. If the corporate
database facilitates data retrieval, it may even make a positive contribution towards improved
infor
mation for management.



(b)

Design Principle 2

(the corporate database) and
Objective 2

(To provide external users
with information useful for predicting the future cash flows and the earning power of a
firm).


To protect themselves from unwanted scrutiny
, management is unlikely to make its corporate
database available to external users. However, as will be shown in subsequent chapters, such a
database could be used as Sorter and McCarthy have suggested, to provide alternative
accounting income and wealth

measures that are not readily available from present
-
day
systems. At present, journal entries are generated by feeder systems and fed into general ledger
systems to generate a single conventional set of GAAP
-
style accounting reports. The emphasis
in mos
t general ledger systems today (Section 4.2) is on classifying information at the most
detailed cost centre, then aggregating to any required level. Actual figures, month by month,
are compared with budgets, which according to Seed [1981] are likely to in
clude expectations




17

Centre for Research in Security Prices, sponsored by Merrill Lynch.





6
-

36

of inflation
18
. As presently structured, alternative views of those data are very difficult to
generate. Of the software reviewed in Chapter 4, only the Fixed Asset package (Section 4.2.5)
generated alternative figures, i.e., CCA fixed
asset values.


Under Design Principle 1, future accounting systems would be required to generate inflation
-
adjusted figures. To do so they will need access to reliable data. It follows that since Design
Principle 2 could assist with provision of that da
ta, it could make a positive contribution
towards Objective 2.


(c)

Design Principle 2

(the corporate database) and
Objective 3

(To provide accountability
information useful for contract monitoring).


A full transaction history of economically significan
t transactions is required for accountability
reporting. Such histories are already maintained today for company
-
law requirements, though
usually on paper or microfilm. As storage costs of machine
-
readable media, relative to paper,
continue to fall, main
taining those records in machine
-
readable form seems an obvious next
step. The only advantage, in terms of accountability, of storing details of all transactions in
some corporate database would be the possibility of generating some reports that might
oth
erwise not be generated. In short, Design Principle 2 does little to assist with Objective 3,
except indirectly through inflation
-
adjusted reports for contract monitoring in times of
inflation. It is, however, not inconsistent with Objective 3.





18

Q.15 of Seed's questionnaire [1981, p.235] suggests that 58% of 282 respondents to his NAA
-
sponsored study of Fortune 1,000 companies used profit plans in

nominal dollars adjusted for
inflation. Another 23% used profit plans in constant dollars. Only 7.4% of respondents reported that
inflation was "not considered in the planning process". (But Seed's response rate was only 28%, so one
wonders what the pr
oportions would have been for non
-
respondents!)






Figure
6.1: Exhibit II from the Study Group on Business Income's Report

[1952, p.145] (referenced in the quotation on the previous page)





Databases


from
:


subsystem 1
───
>
─────────────


Inflation
-

subsystem 2
───
>
─────────────

Adjusted


: HOW?
───────
>
──

Accounting


:

reports

subsystem n
───
>
─────────────




special information

e.g., replacement costs
──
>
──




Figure 6.3(a) The design problem





Databases

from
:


subsystem 1
──
> interface 1
─────┐





subsystem 2
──
> interface 2
─────┤

J/Es GAAP


:
├────────
>
──

General
──
>


Accounting


:


ledger reports

subsystem n
──
> interface

n
─────┤






journal entries keyed
-
in
───
>
────┘




Figure 6.3(b) The present solution






Databases



from
:



su
bsystem 1

>




├───────────────────────┐

Formula Inflation
-


subsystem 2

>



FAJEs Accounting Adjusted


:


├───
>
───
General
──
>


Accounting


:



REE


Ledger reports


:
├─

data

>


Interpreter
─┘








subsystem n

>


model






















special information






e.g., replacement costs
───
>
────┴────────────────┘



price index series




Figure 6.3(c) Proposed Architecture for Computer
-
based Accounting